Cancer begins with mutations of the genome that can include substitutions, indels, duplications, gene fusions, and genome rearrangements. These mutational signatures vary widely between tissues and, often, between individual tumors. To understand this mutational landscape, cancer researchers need tools for visualizing the structural variation in cancer DNA, and its impact on gene transcription and regulation. For greater insight, these data should be augmented with gene annotations, results from functional genomics project like ENCODE, and population-level data. The classic tool for visualizing human gene annotations is the web-based genome browser. However, genome browsers are not generally well-suited to visualizing large-scale structural variation: for that, circular visualization tools (such as Circos) are often used, but those tools are not very interactive (or web-based). Additionally, few genome browsers allow analysis to be performed within the application itself. JBrowse is the most popular genome browser built using JavaScript and dynamic web technology. Originally aimed at model organism genome projects, it is the web-based browser predominantly used by such projects, and yet its single largest user is now the UK-based Cancer Genome Project. We have developed prototypes of an interactive circular view for JBrowse, a data loading pipeline to import human genome annotations, a framework for running interactive analyses from within the browser. We plan to extend these prototypes to a fully-fledged system for visualizing structural variants and other mutations in cancer genomes, and running analyses on those data. We will integrate this software with the systems of the Cancer Genome Project, the International Cancer Genome Consortium, and the Genomic Data Commons.